Embedded texts

text div body

The same criteria that help you determine what should be encoded as a text rather than a div will help you decide whether or not to use text in its third possible position: as a descendant of body. At first glance it may seem strange that an element which is obligatory at the highest levels of the TEI.2 document structure and which has the vital role of distinguishing things from groups of sub-things in a project’s primary content should also be available for use down in the bowels of a TEI.2 document. And indeed, one can make the case for a principled approach of never using text in this position. The argument here would be that because there must be at least one text above body but need be none below, text really identifies a fundamental unit (or a grouping thereof), and given that an atom cannot exist inside another atom, there cannot be occasion to have text deeply embedded inside a high-level text.

The TEI’s rationale for allowing the third position is that inside primary materials below the level of body you encounter pieces of text that—considered by themselves, out of context—would qualify as things. These pieces of text include poems, orations, short plays, letters, bills, petitions, songs, and so on. However, there are tags available at this level such as div, ab, quote, q, lg, and others, that could be used for these pieces of text, so under what circumstances should one use text?

One policy might be to say that any textual feature inside body that matches the project’s criteria for things should be considered an ‘example of the kind’ and encoded as a text. For example, if you are encoding a story cycle like the Arabian Nights that has narratives embedded deep within other narratives and if you equate a narrative with a thing, then you will probably treat those embedded narratives as embedded texts. Or, you may decide that any utterance known to exist in public record (whether in print, on tape, or on film) qualifies as a thing and therefore if that utterance is reproduced in its entirety within the body of a TEI.2 then it should be tagged with text, on a principle of acknowledging this feature’s independent existence outside of the document it happens to be in.

Several principled, defensible approaches are open to you. What you should not do is use text as a convenience when you can’t quite make other tags work there and the DTD happens to allow text in that position.